As OpenAI restructures, Microsoft locks in long-term gains

Hello, Beatrice Nolan here, replacing Jeremy Cahn, who returned today from the Fortune Global Forum in Riyadh. In big AI news, OpenAI and Microsoft announced that they have reached an agreement on the future of their partnership that allows OpenAI to complete its long-awaited corporate restructuring.

The arrangement turns OpenAI’s former profit-seeking arm into a public benefit corporation that can issue traditional stock and will give shareholders a more prominent voice in OpenAI’s governance — two changes that were considered crucial for OpenAI to continue raising the billions of dollars in capital it will need to build more advanced AI models, build massive data centers, and continue its push to become… A major technology platform for consumers and businesses.

Under the deal, the new OpenAI Group PBC will remain under the control of the non-profit OpenAI Foundation. The nearly year-long negotiations with Microsoft, which caused significant tension between the two companies, were the main hurdle for OpenAI to complete the restructuring. At first glance, it appears that Microsoft has extracted major concessions from the AI ​​lab.

The tech giant – which pumped out more than $13 billion In OpenAI since 2019 – will acquire a 27% stake in OpenAI. Given OpenAI’s current valuation, Microsoft’s position is worth around $135 billion. It will also retain access to OpenAI technology until 2032, including any models that reach the Artificial General Intelligence (AGI) stage. Whether the models have crossed that threshold — which OpenAI has previously publicly defined as an AI system capable of performing most economically valuable cognitive tasks as well as or better than a human — will now be verified by an independent panel of experts.

Previously, OpenAI alone could decide when an AGI was accessed, which was seen as a potential leverage point to terminate or change Microsoft’s rights under their partnership. The arrangement is said to have raised tensions, with Microsoft reportedly concerned that OpenAI could prematurely announce general artificial intelligence, using a high-performance AI model as a milestone, which would have significant financial and intellectual property implications for both companies. So this independent panel decision is a win for Microsoft.

The deal also allows Microsoft to pursue AGI independently, or with third parties, while still requiring OpenAI to share many of its research technologies and discoveries. Under the new deal, Microsoft will retain access to much of OpenAI’s core research methods and systems, though the company will not have access to consumer OpenAI hardware, or the model weights and underlying architectural details of any models deemed “research.” (It will retain the rights to these key technical details of OpenAI’s production models until 2032.)

However, this gives Microsoft visibility into things like OpenAI’s model training infrastructure and optimization methods, along with the opportunity to take what it’s learned from OpenAI’s research methods and apply that knowledge to developing its own AGI models. Complicating this further is the expansion of Microsoft’s partnership with Anthropic, with Claude now available in Microsoft 365 and Excel.

Microsoft has given up its cloud exclusivity with OpenAI, which technology analyst Zeus Keravala described as a “major concession on its part.” However, he also noted that the company received several significant structural concessions from OpenAI in return that exceeded this amount.

He said: “These privileges ensure the continuity and value of Microsoft’s investments.” luck. “Essentially, Microsoft has traded the exclusivity of cloud computing, something it was struggling to achieve anyway, for technological certainty and long-term access to IP.”

Investors seemed to agree as the company’s shares rose 2%, pushing its market value above $4 trillion again the day before the company reported its third-quarter earnings.

Regulatory win for OpenAI

For its part, OpenAI will now have access to the full funding promised by investors, including SoftBank, Thrive and other venture capital firms, that were conditional on the restructuring. The move also enables OpenAI to more easily raise additional capital in the future.

The company also appears to have cleared a critical regulatory hurdle. Following news of the deal, Delaware Attorney General Cathy Jennings announced that her office had issued a “statement of no objection” to the proposed corporate recapitalization.

California Attorney General Rob Bonta said luck In a statement, it obtained “secured privileges ensuring that the charitable assets will be used for their intended purpose, safety will be prioritized, as well as a commitment that OpenAI will remain here in California.” As a result, Bonta said his office “will not be in court to oppose OpenAI’s recapitalization plan.”

This is a blow to many nonprofits that have been campaigning against the restructuring, arguing that OpenAI has deviated from its core mission of developing general artificial intelligence in a way that “benefits all of humanity” and that it has prioritized shipping products over the safety of the AI.

These non-profit groups have been lobbying district attorney’s offices to block the deal. Advocacy groups, nonprofits, and some former OpenAI employees, as well as OpenAI co-founder-turned-bitter business competitor Elon Musk, have publicly opposed the restructuring for various reasons. Musk said the move was evidence that OpenAI CEO Sam Altman and co-founder Greg Brockman deceived him when they made the initial multi-million dollar donations to set up the lab. Others claimed that the restructuring risks funneling profits that should have had a public charitable purpose into the pockets of investors in OpenAI projects, co-founders, and employees.

OpenAI has tried to mitigate some of these objections by pledging that its nonprofit OpenAI Foundation will make significant grants for charitable purposes, including a just-announced commitment of $25 billion to projects aimed at improving health, treating disease or aiming to increase society’s resilience to some of the disruptions AI is likely to cause, including the potential for widespread job loss.

However, with regulatory approval gained and Microsoft’s concerns addressed, OpenAI finally has its own profit structure — and billions of dollars in capital. Microsoft is keeping most of what it found valuable in the OpenAI partnership, while working to secure a significant equity stake that could generate a windfall for its shareholders.

With that said, here’s more AI news.

Beatrice Nolan

bea.nolan@fortune.com

Fortune on artificial intelligence

Andreessen Horowitz partner says open source AI is ‘China’s toy now’ — and that’s a problem for the US and its allies — Beatrice Nolan

Now we know that AI won’t take all our jobs, Silicon Valley must fix its fundamental mistake: the theater of automation must end. Joel Haroun

Qualcomm CEO warns that ‘everyone is playing to win’ when it comes to the AI ​​bubble, but it’s still too early to tell who will succeed — Beatrice Nolan

After Microsoft invested $13 billion in OpenAI, its AI head criticized erotica features like ChatGPT: ‘This is very dangerous’ – Sasha Rogelberg

Eye on artificial intelligence news

Elon Musk launches Grokipedia as a new competitor to Wikipedia. Elon Musk promoted his new project, Grokipediaan encyclopedia based on artificial intelligence, as an unbiased alternative to Wikipedia. The first pages are similar to the format of Wikipedia, but according to washington post, It is presented in a more right-leaning tone. The outlet also identified several factual errors. Musk has raised issues with Wikipedia before, speaking publicly about what he sees as the site’s shift to the left. The site, which launched this week, currently hosts about 885,000 articles, far fewer than Wikipedia’s more than 8 million articles. Read more at The Washington Post.

Claude adds Excel integration and real-time market data tools. Anthropic has launched a finance-focused upgrade to its Claude chatbot, adding direct integration with Microsoft Excel and seven new real-time data connectors. Anthropic’s Claude can now analyse, edit, and create spreadsheets directly within Excel, providing financial professionals with a more interactive way to work with data. The new links connect Claude to major financial platforms – including Moody’s, the London Stock Exchange Group, and MT Newswires – allowing him to access live market updates, earnings call transcripts, and investment research. Anthropic has also introduced six new “Agent Skills” designed specifically for finance, enabling Claude to produce reports, model cash flows, and automatically create company profiles. Read more here.

Qualcomm is entering the race for artificial intelligence chips with data center processors. Qualcomm has announced a big move into the AI ​​data center market with two new accelerator chips designed to challenge the dominance of Nvidia and AMD. The chips, designed to run AI inference rather than training, can fill an entire liquid-cooled server rack, and are built on Qualcomm’s Hexagon neural processing units, which are used in its smartphone chips. With demand for AI computing expected to drive up to $6.7 trillion in data center spending through 2030, Qualcomm’s entry represents a major expansion beyond its traditional mobile chip business. Qualcomm shares rose 11% after this announcement. Read more at CNBC.

OpenAI urges the United States to boost energy production to stay ahead in the AI ​​race. OpenAI is calling on the US government to dramatically expand national energy production, warning that the country risks falling behind China in the global AI race without massive new energy investments. In an 11-page report to the White House Office of Science and Technology Policy, the company urged the United States to commit to building 100 gigawatts of new energy capacity each year — nearly double what the country added in 2024. OpenAI noted that China added 429 gigawatts last year compared to 51 gigawatts in the United States, creating what it called an emerging “electron gap.” Read more at CNBC.

Eye on artificial intelligence research

Another MIT report says the benefits of AI may not be as clear as expected. This time in coding. A new MIT Technology Review Insights report, conducted in collaboration with Snowflake, found that 77% of data engineers face higher workloads despite the widespread adoption of AI tools aimed at boosting productivity. The survey of 400 senior technology executives found that 83% of organizations have already deployed AI-based data engineering tools, but 45% cite integration complexity, while 38% report breadth and fragmentation of tools as key adoption challenges. While AI is automating many data tasks, the proliferation of discrete systems has created a productivity paradox where individual tasks are faster, but the overall workflow is slower, according to the report. Data engineers now spend 37% of their time on AI-related projects, compared to 19% two years ago, and they expect this percentage to reach 61% within two years.

Artificial Intelligence Evaluation

November 10-13: Web Summit, Lisbon.

November 26-27: World Conference on Artificial Intelligence, London.

December 2-7: NewReps, San Diego.

December 8-9: Brainstorming Fortune AI San Francisco. Apply to attend here.

Brain food

Do AI web browsers pose a security risk? After OpenAI launched its long-awaited web browser last week, I wrote about some of the risks of rapid injection. Since then, perhaps not surprisingly, more security risks have emerged. Security firm LayerX has discovered a potentially major vulnerability in the ChatGPT Atlas browser that allows attackers to inject malicious instructions into ChatGPT memory. Using a cross-site request forgery (CSRF) attack, hackers can exploit a logged in user’s session to plant hidden instructions that persist across devices and browsers. Once infected, ChatGPT may inadvertently execute these instructions, potentially spreading malware. LayerX tests showed that ChatGPT Atlas was particularly vulnerable, blocking only about 5.8% of the phishing attacks they tested, leaving users up to 90% more vulnerable than Chrome or Edge, which blocked nearly half of these attacks. Cybersecurity is always a game of cat and mouse, with companies identifying security flaws and then patching them, but the scale of the security risks begs the question of whether AI browsers are too risky to be trusted with the kind of deep system access they require to be useful.